Note: Descriptions are shown in the official language in which they were submitted.
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
REVERSE LINK CORRELATION FILTER
IN WIRELESS COMMUNICATION SYSTEMS
RELATED APPLICATIONS
The present application claims priority to U.S. Provisional Patent
Application Serial No. 60/184,364 entitled REVERSE LINK CORRELATION FILTER
IN WIRELESS COMMUNICATION SYSTEMS filed on February 23, 2000. Copending
U.S. Patent Application Serial No. 09/497,440 filed on February 3, 2000 and
entitled
PILOT SYMBOL ASSISTED MODULATION AND DEMODULATION IN
WIRELESS COMMUNICATION SYSTEMS, is also a related application.
FIELD OF THE INVENTION
The present invention relates to a code division multiple access (CDMA)
communication system and, more particularly to a receiver having a correlation
filter in
such a CDMA communication system.
DESCRIPTION OF THE RELATED ART
Code Division Multiple Access (CDMA) modulation, which is known in
-1-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
the art, is a multi-user access transmission scheme in which signals from
different users
overlap both in frequency and in time. This is in contrast to Frequency
Division Multiple
Access (FDMA), also known in the art, in which user signals overlap in time,
but are
assigned unique frequencies, and Time Division Multiple Access (TDMA) in which
user
signals overlap in frequency, but are assigned unique time slots. CDMA
signaling is
frequently used in cellular communication systems between a base station (BS)
within
a cell and a plurality of mobile stations (MS) in the possession of users
within the cell.
The CDMA transmitted signal for each user that broadcast from the user's
mobile station
(MS) is spread over a wide bandwidth, which is greater than the initial user
information
bandwidth. Each user's signal is spread by a different spreading code to
create a
wideband spread. All of the spread wideband signals transmitted by the
different users
are received at the base station (BS) and form a composite received signal.
The receiver
at the base station (BS) distinguishes different users by using a local copy
(or local
reference) of the spreading code, which is available to both the mobile
stations and the
base station in the CDMA system. Such a process is called channelization. In
an
exemplary CDMA system according to the IS-95 standard and which is well known
in
the art, channelization in the reverse link, is e., when a mobile station (MS)
is transmitting
to a base station (BS) in the system, is accomplished using a wideband code
called a
pseudorandom noise (PN) code, also known in the art. The receiver at the base
station
(BS) sifts the desired signal from a particular user out of the composite
signal by
correlating, i.e., using a correlation filter (CF), on the composite signal
with the original
wideband code. All other signals having codes that do not match the code for
the desired
user code are rejected.
An exemplary CDMA wireless system includes a plurality of data
channels, e.g., the access and traffic channels (and more channels depending
on the
design of the CDMA system). In the reverse link, the traffic channel is used
to transmit
user data and voice, as well as signaling messages. The access channel is used
by the
mobile station (MS), e.g., a cellular phone, to communicate control
information with the
base station (BS) in the wireless system when the MS does not have a traffic
channel
assigned. In particular, the MS uses the access channel to make call
originations and to
respond to pages and orders. These data channels in the CDMA system have
different
-2-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
functions and data rates. A receiver in the MS designed to accommodate data
transmission in the different channels requires various types of correlation
filter (CF) and
digital signal processing (DSP) designs for different data rates. Such
requirements
contribute to the complexity and increase the cost of the receiver design.
There is, therefore, a general need in the art for a wireless system with a
flexible, non-complex receiver design. A wireless system is particularly
needed that
provides a single correlation filter (CF) in the receiver which can be used in
receiving
data in all data channels. There is a further need for a receiver design with
a correlation
filter that serves all data channels and is also DSP programmable, which
enhances system
flexibility.
SUMMARY OF THE INVENTION
The invention relates to a reverse link receiver in wireless systems and a
correlation filter thereof. A transmitter and a receiver are provided in the
reverse link of
a wireless system according to the invention. The receiver includes (1) a
field
programmable gate array (FPGA) which comprises a pseudorandom noise (PN) code
generator, (2) a pilot post processor, (3) a data post processor (4) a
correlation filter (CF),
and (5) an external programmable processor such as a digital signal processor
(DSP).
The field programmable gate array (FPGA) and the correlation filter (CF),
along with the
digital signal processor (DSP), are included in the receiver to recover the
original data
transmitted by the transmitter. The correlation filter (CF) comprises a
(correlating filter)
CF core for processing data in the three channels, namely the access,
maintenance and
traffic channels. The external programmable processor is used to control and
post-
process the outputs of the field programmable gate array FPGA. A mode
controller in
the external programmable processor controls the channel selection (from the
access,
maintenance and traffic channels) and the channel symbol mode (data and/or
pilot). The
pilot post processor and data post processor in the FPGA, in conjunction with
the external
programmable processor, provide pilot symbol aided QPSK demodulation of up to
3
multipaths received at the receiver. QPSK modulation is a modulation technique
that
allows the transmission of two bits of information in each symbol period. QPSK
modulation makes use of the quadrature component I in addition to the in-phase
-3-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
component Q of a symbol in the frame being transmitted from the transmitter to
the
receiver. The I and Q components are typically viewed as the real and
imaginary parts
of a complex signal being transmitted in the channels of the CDMA system. In
QPSK,
the in-phase component, I, and the quadrature component, Q, can be combined
without
interfering with each other (i.e., they are orthogonal to each other) which
doubles the
bandwidth efficiency in comparison with simply transmitting one bit of
information in
a symbol period. Using time multiplexing, the CF core provides pilot symbol
correlation
at the three data or chip rates (tiers 1, 2 and 3). A chip is a unit of time
which
corresponds to the output interval of the PN spreading code. The chip time
determines
the bandwidth of the CDMA waveform and the chip time divided by the user
symbol
time determines the spreading factor of the system. For example, the sampling
period for
a chip in CDMA standard IS-95, known in the art, is 1 seconds. The pilot post
1228800
processing and data post processing in the FPGA, in conjunction with the data
post
processor and pilot post processor in the external programmable processor
(such as a
DSP), provide QPSK demodulation and recovery of the original data transmitted
by the
transmitter for all three channels, i. e., the access, maintenance and traffic
channels. The
CF core according to the invention performs 8-chip (i.e., tier 1 rate or the
basic
despreading rate) complex correlation with 64 correlation lags while allowing
no data
loss as the 64 correlation lags are being processed, where a lag is a time
instant for which
the PN code (or orthogonal code, Walsh code) is held constant so that outputs
can be
generated. The basic despreading rate can also be 4 chips per symbol. This is
advantageously achieved with a single time-multiplexed 8-chip correlator
engine (i. e., CF
core). The CF core, which is an 8-chip correlation engine, is time multiplexed
to allow
multiple 8-chip correlations to be performed with the same correlation engine.
Furthermore, the invention advantageously provides the ability to generate
integer
multiples of the 8-chip correlation for larger correlation lengths, e.g., 32
chips (tier 2) or
128 chips (tier 3). The CF design according to the invention, in conjunction
with the
DSP, provides temporal diversity of the data signals by combining CF outputs.
Spatial
diversity is also achieved by providing a plurality of correlation filters in
accordance with
the CF design of the invention. Diversity is a technique employed to avoid or
mitigate
the negative effects of fading and interference. Diversity generally refers to
the ability
-4-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
of a communication system to receive data or information via several
independently
fading channels. In general, diversity enhances a receiver's ability to
combine or select
(or both) data signals arriving from these independently fading channels,
thereby enabling
(or facilitating) the extraction of data channels. A particular type of
diversity is temporal
or time diversity, where the same data signals transmitted in different
multipaths and
received at different time points at the receiver provide the diversity needed
for
combining or selecting the data signals. An exemplary diversity technique is
maximum
ratio combining, or MRC, known in the art. MRC provides sequences of weights
in the
multitude of data channels in the communication system. A sequence of distinct
weights
is assigned to segments of a data signal being transmitted. Copies of the data
signal are
produced for the antennas used in transmitting the data signal. This gives
rise to temporal
diversity at the receiver when copies of the data signal are received. With a
single,
common correlation filter design, the invention advantageously provides
temporal
diversity for all the data channels and modes without the necessity of using
different
correlation filters for different modes or channels. Temporal diversity is
achieved by
providing three outputs from the different multipaths and combining the three
outputs
into one, e.g., using MRC.
According to an embodiment of the method of the invention, a single,
common correlation filter (CF) design is provided in a wireless system using
CDMA.
A plurality of channels with different data rates are provided in the wireless
system. The
channels provided in the wireless system include the access channel, the
maintenance
channel, and the traffic channel in which information (e.g., pilot or data
symbols or both)
is transmitted at the tier 1 (which is the basic despreading rate), tier 2 and
tier 3 rates.
The data rate for transmitting the information is programmable by the external
programmable processor. A user-unique code, such as a PN code (or orthogonal
code,
Walsh code), is applied to the information being transmitted in the channels
of the
wireless system. The information is QPSK modulated and transmitted in any one
of the
channels and at any data rate. The transmitted information is correlated at
the basic
despreading rate (i.e., the tier 1 rate) using time multiplexing in the
correlation filter (CF)
of the wireless system. The basic despreading rate can be 4 or 8 chips per
symbol. The
correlated information is then demultiplexed and QPSK-demodulated. The
demodulated
-5-
CA 02400934 2009-09-25
information is summed at the proper integer multiple of the tier 1 rate (the
basic
despreading rate) to achieve the tier 2 and tier 3 rates if needed. One or
more
signal components (in terms of the received power, signal-to-noise ratio or
multipath width) are selected in a window or time period for optimal
information recovery. The selecting step can also be implemented according to
a
preprogrammed time alignment. Furthermore, three outputs from the
demodulated information can be provided and combined for temporal diversity.
Spatial diversity is achieved by providing the single, common correlation
filter
design in a plurality of receivers in the wireless system. All the process
steps
according to the invention described herein are advantageously accomplished
using a single, common correlation filter (CF) design, which eliminates the
need
for additional correlators or correlation filters for processing received
information having multiple data rates.
According to an embodiment of the present disclosure there
is provided a method for use in wireless communication. The method comprises:
receiving signals over a code division multiple access (CDMA) channel, each
received signal having a user-unique code and an associated data rate;
correlating the received signal with the user-unique code at a basic data rate
using time multiplexing in a single correlation filter to produce one or more
sets
of correlated outputs; demultiplexing and demodulating the one or more sets of
correlated outputs to generate demultiplexed and demodulated sets of
correlated
outputs; and on a condition that the associated data rate is a multiple of the
basic data rate, summing the demultiplexed and demodulated sets of correlated
outputs at the multiple of the basic data rate to recover user information
from
the received signal at the associated data rate.
According to another embodiment of the present disclosure
there is provided an apparatus for use in wireless communications. The
apparatus comprises: a receiver configured to receive signals over a code
division
multiple access (CDMA) channel, each received signal having a user-unique code
and an associated data rate; a correlation filter configured to correlate the
-6-
CA 02400934 2009-09-25
received signal with the user-unique code at a basic data rate using time
multiplexing to produce one or more sets of correlated outputs; a
demultiplexer
configured to demultiplex the one or more sets of correlated outputs to
generate
demultiplexed sets of correlated outputs; and a data post processor. The data
post processor is configured to: demodulate the sets of demultiplexed outputs
to
generate demultiplexed and demodulated sets of correlated outputs; and on a
condition that the associated data rate is a multiple of the basic data rate,
sum
the demultiplexed and demodulated sets of correlated outputs at the multiple
of
the basic data rate to recover user information from the received signal at
the
associated data rate.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other features, aspects, and advantages of the invention
will become more readily apparent with reference to the following detailed
description of a presently preferred, but nonetheless illustrative, embodiment
when read in conjunction with the accompanying drawings. In the
accompanying drawings:
Figure 1 is a diagram generally illustrating a transmitter and a
receiver in the reverse link in accordance with the invention;
Figure 2 is a diagram illustrating the structure of a frame of
data/pilot symbols being transmitted in various channels in an embodiment
according to the invention;
Figure 3 is a diagram illustrating a receiver with a field
programmable gate array (FPGA) forming a correlation filter (CF) according to
the invention;
Figure 4 is a diagram illustrating an exemplary field
programmable gate array forming a correlation filter (CF) core for all
channels
in the wireless system according to the invention;
Figure 5 is a diagram illustrating an exemplary 8-chip correlation
with a pseudorandom noise (PN) code according to the invention;
-6a-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
Figure 6 is a diagram illustrating the pilot post processing of all the
channels in an embodiment of the wireless system according to the invention;
Figure 6A shows diagrams illustrating the gain coefficient setting of an
accumulating filtering random access memory (AFRAM) and a magnitude
accumulating
filtering random access memory (MAFRAM) for three different data rates in an
embodiment according to the invention;
Figure 6B is a diagram illustrating the basic structure of an embodiment
of the AFRAM and MAFRAM according to the invention;
Figure 7 is a diagram illustrating the multipath search processing of
multipath response peaks according to the invention;
Figure 7A is a flow diagram illustrating an embodiment of the method of
the multipath search processing of multipath response peaks according to the
invention;
Figure 8 is a diagram illustrating the data post processing in the channels
of the wireless system according to the invention; and
Figure 9 is a flow diagram illustrating an embodiment of the method of
QPSK modulation according to the invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Turning to Figure 1, a transmitter 10 and a receiver 20 communicate over
the reverse link in an exemplary wireless system according to the invention.
In Figure
1, transmitter 10, Tx 1, is a part of a mobile station (MS) or mobile cellular
phone
communicating with receiver 20, Rx 2, which is part of a base station (BS).
Digital
information at a basic data rate is coded or spread to a transmit data rate
(or chip rate) at
transmitter 10. A user-unique digital code (the signature or spreading
sequence) is
applied to the digital information being transmitted, which increases the
bandwidth. The
application of the digital code, such as a pseudorandom noise (PN) code,
typically
involves multiplication or logical XOR (exclusive-or) operations on the
digital
information being transmitted with the PN code (or orthogonal code, Walsh
code). The
resulting transmitted data sequences or chips are then QPSK modulated at
transmitter 10
to generate an output signal. The output signal is added to other similarly
processed
output signals for multi-channel transmission to the receiver 20 over a
communications
-7
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
medium. The output signals of multiple users advantageously share a single
transmission
communications frequency, with the multiple signals appearing to be located on
top of
each other in both the frequency domain and the time domain. Because the
applied
digital codes are orthogonal and user unique, each output signal transmitted
over the
shared communications frequency is similarly unique, and can be distinguished
from each
other through the application of proper processing techniques at receiver 20.
The receiver 20, comprises a correlation filter CF 3 formed by a field
programmable gate array FPGA 5 and a digital signal processor DSP 4. In
receiver 20,
the received signals are QPSK-demodulated and the appropriate digital code for
the user
of interest is applied to (i.e., multiplied with) the signal to despread and
remove the
coding from the desired transmitted signal, and to return it to the basic data
rate. Where
the digital code (i.e., PN code, orthogonal code, or Walsh code) is applied to
other
transmitted and received signals, there is no despreading as the signals
maintain their chip
rate.
This despreading operation effectively comprises a correlation process
comparing the received signal with the appropriate digital code. In accordance
with
QPSK, the transmitted data sequences or chips include an in-phase component
(I) and a
quadrat-Lire phase component (Q), which are the real and imaginary parts of a
complex
signal. The despreading operation performed by receiver 20 correlates the I
and Q
components of the received complex signal to the appropriate digital code or
signature
sequence. This is accomplished using a single correlation filter design (e.g.,
CF 3) for
all data rates in the channels of the wireless system.
In a preferred embodiment of the invention, the exemplary wireless system
provides an access channel 11, a maintenance channel 12, and a traffic channel
13 in the
reverse link. The correlation filter CF 3 is DSP programmable (i. e.,
controlled by DSP
4) and can be configured to perform the digital signal processing needed in
receiving data
from any of the three channels 11, 12 and 13 and at all user data rates. After
processing
in the correlation filter CF 3 and digital signal processor 4, the original
data from the
transmitter 10, is recovered. The system and the various embodiments according
to the
invention are described in further detail below.
Figure 2 illustrates the structure of a frame of data/pilot symbols being
-8-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
transmitted in the access, maintenance and traffic channels in an embodiment
according
to the invention. Inthis particular embodiment, the access channel (e.g.,
access channel
11 of Figure 1) is used by the receiver 20 to communicate with the transmitter
10 when
no traffic channel is. assigned. The access, channel provides a user with a
shared data
channel for requesting access to the CDMA system. The maintenance channel
(e.g.,
maintenance channel 12 of Figure 1) maintains the timing of the transmitter
and the
receiver in the reverse link using a plurality of pilot symbols. No data
symbols are
transmitted in the maintenance channel. The maintenance channel provides a
user with
the ability to maintain synchronization with the CDMA system during periods of
inactivity. User data and signaling messages (i. e., pilot symbols) from the
transmitter are
transmitted to the receiver Rx 2 in the traffic channel (e.g., traffic channel
13 of Figure
1).
The wireless system according to the invention provides three tiers of data
rates, i.e., tier 1 (which is the basic despreading rate), tier 2 and tier 3,
for use by the
channels (access channel 11, maintenance channel 12, and traffic channel 13).
At tier 1,
the transmitter transmits 8 chips per symbol to the receiver. At tier 2, the
transmitter
transmits 32 chips per symbol to the receiver in the reverse link. At tier 3,
the transmitter
transmits 128 chips per symbol in the reverse link of the exemplary wireless
system
according to the invention. Transmitting data in the access, maintenance and
traffic
channels is described in further detail below.
When transmitting a frame of data and/or pilot symbols in the reverse link
using the access channel, the frame comprises a preamble, succeeded by data,
another
preamble, and data (Figure 2). The frame is modulated using QPSK (quadrature
phase
shift keying), which is known in the art, and channelized (spread) using a
unique digital
code, such as a pseudorandom noise (PN) code, also known in the art.
When transmitting the modulated frame inthe access channel, three access
modes of operation for the access channel are provided, namely, the preamble
mode, the
pilot mode, and the data mode. A PN epoch marks the end of one frame and the
beginning of another frame. A preamble of a frame in the access channel
comprises a
plurality of pilot symbols transmitted at the tier 2 rate (i.e., 32 chips per
symbol in the
frame). In the preamble mode, pilot symbols only and no data symbols are
transmitted.
-9-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
Data, which succeed the preamble in the frame being transmitted in the access
channel,
comprise a plurality of data symbols and pilot symbols interleaved at the tier
2 rate (i. e.,
32 chips per symbol). A mode controller in DSP 4 obntrols the mode of
operation of the
access channel and 'switches between.the preamble, pilot and data modes.
When transmitting a fratnein the maintenance channel, the frame includes
information for multiple users denoted P1, P2, P3, ..., P128, and no data
which is time
division multiplexed (TDM) with two pilot symbols per user. The data rate is
Tier 3 with
128 chips per symbol (Figure 2). Only one mode of operation is provided in the
maintenance channel, namely, the pilot mode.
When transmitting a frame in the reverse link of the traffic channel, two
transmission modes are provided, namely, the pilot mode and the data mode. A
frame
comprises pilot and data symbols, where the pilot symbols are placed at
periodic intervals
between the data symbols. The data and pilot symbols may be transmitted at any
data
rate; transmission at tier 1 rate (i. e., 8 chips per symbol), tier 2 rate (32
chips per symbol)
and tier 3 (128 chips per symbol). The data symbols are transmitted at the
same rate as
that of the pilot symbols. Similarly, a mode controller in DSP 4 controls the
mode of
operation of the traffic channel and switches between the pilot mode and the
data mode.
Figure 3 is a diagram illustrating a receiver with field programmable gate
array (FPGA 5) and controlled by a digital signal processor (DSP) 4 according
to the
invention. The FPGA 5 comprises an FPGA pilot post processor 33, an FPGA data
post
processor 35, pseudorandom noise (PN) code generator 46, and a correlation
filter CF 3
having a CF core 31. The field programmable gate array FPGA 5 with the
correlation
filter CF 3, and the digital signal processor DSP 4 are included in the
receiver 20 to
recover the original data transmitted by the transmitter 10. The CF core 31 is
used for
despreading received data in the three channels (access, maintenance and
traffic) 11, 12
and 13. The digital signal processor DSP 4 is used to control and post-process
the
outputs of the field programmable gate array FPGA 5.
DSP 4 includes a mode controller 41 which indicates the channel selection
(from the access, maintenance and traffic channels) for transmitting data
and/or pilot
symbols. DSP 4 also includes a pilot DSP post processor 34 and a DSP data post
processor 36 which provide pilot symbol-aided QPSK Demodulation of the
multipaths
-10-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
received at the receiver 20. Pilot symbol aided demodulation is described in
the
copending U.S. Patent Application Serial No. 09/497,440 entitled PILOT SYMBOL
ASSISTED MODULATION AND DEMODULATION IN WIRELESS
COMMUNICATION SYSTEMS, which is incorporated herein by reference.
Figure 4 is a diagram that illustrates an exemplary field programmable
gate array (FPGA) 5 with a correlation filter (CF) core 31 for all channels in
the wireless
system according to the invention. The CF core according to the invention
performs 8-
chip complex correlation with 64 correlation lags, while allowing no data loss
as the 64
correlation lags are being processed. A correlation lag is' a time instant for
which the PN
code (or orthogonal code, Walsh code) is held constant, so that the received
data in the
receiver 20 is correlated based on the PN code (or orthogonal code, Walsh
code) and
corresponding outputs therefor are generated. Correlation lags are computed
independent
of the channel type and data rate. This is advantageously achieved with a
single time-
multiplexed 8-chip correlator engine (i.e., CF core 31) which allows multiple
8-chip
correlations to be performed with the same correlation engine. Furthermore,
the
invention advantageously provides the ability to generate integer multiples of
the 8-chip
correlation (e.g., 32 or 128 chips) for larger correlation lengths. The
operation of the CF
core according to the invention is described in further detail below.
As seen in Figure 4, the field programmable gate array 5 includes a
correlation filter CF 3 with a CF core 31, a PN (pseudorandom noise) code
generator 46
with correlator pattern buffers 401 through 408, a multiplexer (MUX) 44, a
window
processor 43, a demultiplexer (DEMUX) 45, and symbol processors 411 through
418.
CF core 31 (which is explained in more detail with reference to Figure 5) is a
correlation
engine which is a single 8-chip correlation module that provides despreading
for all
channels (i. e., access, maintenance and traffic channels 11, 12 and 13). CF
core 31 is the
basic correlation engine which can be used to obtain any x-chip correlation, x
being the
factor of 8, of the QPSK-modulated frame of data and/or pilot symbols received
from the
transmitter 10.
PN code generator 46 in the receiver 20 (Rx 2) generates the local PN
reference of the PN code (or orthogonal code, Walsh code) used by the
transmitter 10 (Tx
1). PN code generator 46 is DSP programmable and controlled by DSP 4 (Figure
3) to
-11-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
generate the proper PN code phase. The PN code (or orthogonal code, Walsh
code)
phase used by each user is established during transmitter-receiver connection
setup and
remains fixed for the duration of that user's call, i.e., during that
connection time to the
wireless system.
The PN code phases from the PN code generator 46 are applied to
respective correlator patterns 401 through 408. The outputs or the correlator
patterns are
selectively e.g., sequentially, processed in CF core 31 via the MUX 44. The
length of
each correlator pattern is 8 chips.. In particular, the MUX 44 loads 8 chips
of PN
reference data of the PN code (or orthogonal code, Walsh code) from one of the
correlator patterns into a buffer so that the PN reference can be used by the
time
multiplexed CF core 31 at the proper time. The 8 chips of the PN code (from
respective
correlator patterns 401 through 408) are held in the buffer for a period of 64
chips and are
used to despread the received data from the transmitter 10. CF core 31
generates a
correlation value for each shift of the received waveform from MUX 44 as it
passes the
PN reference. The time multiplexed CF core 31 allows the generation of
multiple
correlation lags with a single correlation engine.
Assuming that the modulated frame received from the transmitter 10 is
oversampled four times the normal sampling rate set for the system, 256
correlation lags
are provided in the 64-chip period. Furthermore, time multiplexing allows a
single 8-
chip correlator engine (e.g., CF core 31) to provide multiple correlation lags
with no loss
of data. This process is repeated for the next 8 chips of the PN code (or
orthogonal code,
Walsh code).
The correlation core 31 performs a complex correlation. That is, each of
the CF core 31 outputs one of four real correlation values. The correlation
values are the
results of an 8-chip complex correlation which is broken into four real 8-chip
correlations
in the correlation engine. The result of the four real correlations represents
four real
multiplications in a complex multiplication, as follows:
(a+jb)*(c+jd) = ac - bd + jbc + jad (Eq. 1)
where ac = II, bd = QQ, be = QI and ad = IQ.
The 8-chip complex correlation corresponds to the smallest despreading
factor used in the wireless system according to the invention, which is 8
chips for a tier
-12-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
1 rate (the basic despreading rate). All other data rates in the wireless
system are
multiples of 8 and can therefore be generated by summing multiple 8-chip
correlation
outputs. If the number of correlation lags is to be greater than 8 chips
(i.e., the PN code
(or orthogonal code, Walsh code) is being held constant for more than 8
chips), the next
8-chip time, period will have passed and the correlation on the following data
will not be
properly calculated. As a result, more than one pattern is needed and the
number of
correlators will be a function of the number of correlation lags needed.
Figure 5 is a timing diagram illustrating an exemplary time multiplexed,
8-chip complex correlation with a pseudorandom noise (PN) code according to
the
invention. Since the number of correlation lags in the wireless system
according to the
invention is established at 64 chips, there is provided a methodology for
performing
correlations in parallel with time shifting to properly correlate the next 56
chips. As a
result, 8-chip time multiplexed correlations are staggered in time by 8 chips.
That is, for
every 8 chips a new correlation begins while the preceding correlation is
still being
performed. Therefore, if the correlation output of the 8-chip complex
correlation in CF
core 31 is to have 64 correlation lags, eight different 8-chip correlator
patterns (e.g., 401
through 408 of Figure 4) are processed, each spaced by 8 chips so that no data
is missed,
as shown in Figure 5. Each of the eight phases of the correlator pattern
(e.g., Corr Phase
1 through 8) is shifted by 8 chips, which is shown by the load notation of LD
in Figure
5. Once the 8 chips of the PN code (e.g., from PN code generator 46) have been
loaded,
each phase of the correlator pattern produces 64 chips of correlation lag for
that set of 8
PN chips in a store operation denoted ST in Figure 5. To avoid missing the
next 56 chips
of data, seven other phases of the correlator pattern are set up with the next
56 chips of
the PN code and 64 chips of correlation lag calculated for the 8 chips of the
PN code.
During each correlator phase (e.g., Corr Phase 1 through 8), CF core 31
generates
correlation lags for a tier 1 symbol. Each phase of the 64 correlation lags of
valid data
is shifted in time, as noted in Figure 5, e.g., Phase 1 Data Potentially
Valid, Phase 2 Data
Potentially Valid. The 8-chip time multiplexed correlation engine (e.g., CF
core 31)
produces the 4 real components of a complex correlation for a tier 1 symbol,
each having
64 chips of correlation lag. The time multiplexing cycle for CF core 31 is 8
symbols in
length, where 8 symbols of tier 1 data are produced per cycle.
-13-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
The window processor 43 of Figure 4, the demultiplexer (DEMUX) 45,
and the symbol processors 411 through 418 together demodulate the received
data from
the transmitter 10 using QPSK demodulation. An exemplary QPSK demodulation is
described herein and'in the copending U.S. Patent Application Serial No.
09/497,440
filed on February 3, 2000 and entitled PILOT SYMBOL ASSISTED MODULATION
AND DEMODULATION IN WIRELESS COMMUNICATION SYSTEMS. The
window processor 43, performs a phase de-rotation of the received signal with
a channel
estimate from the pilot post-processing (Figure 3). In particular, the window
processor
43 performs complex multiplication of the correlation values corresponding to
each tier
1 symbol from the CF core, and outputs the pilot post processor signal which
estimates
the channel.
The DEMUX 45'takes the output of the time multiplexed CF core 31 and
window processor 43 and produces 64 chips of correlation lag for each tier 1
symbol and
routes them to their corresponding symbol processors (411 through 418). There
are eight
symbol processors (411 through 418), one for each tier 1 symbol that is
produced by CF
core 31 during the eight phases of the 8-chip complex correlation. The symbol
processors (411 through 418) are programmed by DSP 4 to select the proper
multipaths
from the output of the window processor 43. The symbol processors take the DSP
programmable number of correlation lags associated with up to three
multipaths, and sum
the correlation lags to form three outputs (Figures 3 and 4). The outputs from
the symbol
processors are always at the tier 1 rate (the basic despreading rate) from
which the DSP
4 can accumulate the outputs to achieve the tier 2 and tier 3 rates. In
addition, for the tier
2 rate (32 chips), four outputs from the symbol processors at the tier 1 rate
(which is the
basic despreading rate) are summed. For the tier 3 rate (128 chips), sixteen
outputs from
the symbol processors at the tier 1 rate are summed.
Figure 6 is a diagram illustrating the pilot post processing of any channel
of the wireless system according to the invention, namely the access channel,
traffic
channel and maintenance channel. In each of the three channel types, a portion
of the
information being transmitted includes pilot symbols. A pilot symbol is a
known
constant value which the receiver 20 uses to estimate the channel conditions
for each
received multipath. As the mode controller 41 of DSP 4 (Figure 3) indicates
pilot
-14-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
processing in the access, traffic or maintenance channel, pilot post processor
33 in FPGA
of the receiver 20 (Rx 2) starts the processing of the pilot symbols
transmitted in the
channel. Two accumulating filtering random access memories (AFRAM) 61 and 63
(shown in Figure 6) having a one-pole infinite impulse response (IIR) filter
are provided
5 in pilot post processing 33 for integration of the pilot symbols being
transmitted in the
channel. Based on the pilot symbols and the IIR filtering thereof, an estimate
of the
channel is obtained for completing pilot symbol aided QPSK demodulation. Each
AFRAM contains 64 correlation lags of IIR-filtered pilot symbols at 4
samples/chip.
This allows receiver 20 to have 64 chips of delay-spread range for searching
the received
multipaths at a resolution of 1/4 of a chip time.
AFRAM 61 (and similarly, AFRAM 63) can act as a one-pole IIR filter
with coefficients a and R. An IIR filter, known in the art, is a digital
filter that linearly
processes sampled data in a recursive manner: That is, an IIR filter samples
continuous-
time data signals with a fixed periodicity and linearly manipulates and
transforms the
samples. The one-pole IIR filter has two coefficients, a and 0, which controls
the
function of the AFRAM. The coefficients (a and (3) are DSP programmable and
controlled by DSP 4. Based on the selection of coefficients, the AFRAM (61 or
63) can
serve three functions, namely storage (or more particularly, random access
memory or
RAM), accumulation, and filtering. As a RAM, the AFRAM stores 256 correlation
lags
of data for a tier 1 pilot symbol. As an accumulator, the AFRAM accumulates
correlation
data over multiple tier 1 pilot symbols to generate tier 2 or tier 3 data. As
a filter, the
AFRAM is a one-pole IIR filter that filters data at the tier 1, 2 or 3 data
rates for pilot
symbol estimation in the presence of noise and interference in the channel.
When the (3
coefficient of the IIR filter equals zero, there is no recursive feedback in
the IIR filter, i. e.,
no filtering, and AFRAM 61 (and similarly AFRAM 63) acts as a sample RAM
(Figure
6) for storing the inputs from summer 621 (or summer 623 for AFRAM 63). If the
(3
coefficient equals one, AFRAM 61 (and similarly AFRAM 63) acts as a simple
accumulator for accumulating the inputs from summer 621 (or summer 623 for
AFRAM
63). If the (3 coefficient is between zero and one, then AFRAM 61 (and
similarly
AFRAM 63) acts as a filter. In effect, by controlling the coefficients (e.g.,
through DSP
4) the AFRAM 61 (and similarly AFRAM 63) can act as a simple RAM for storing
data,
-15-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
an accumulator for accumulating data inputs, or a filter for filtering data
signals. This
allows the system to process data inputs at any of the tier 1, tier 2 and tier
3 rates, and to
integrate the pilot symbols over a time period specified by DSP 4, i. e.,
multiple-tier pilot
symbols at tier 1, tier 2 or tier 3.
Figure 6 also contains a MAFRAM 65, which is a magnitude
accumulating filtering random access memory (MAFRAM 65). MAFRAM 65 also
includes a one-pole IIR filter containing 64 correlation lags of IIR-filtered
pilot symbols.
Similar to AFRAM 61 and AFRAM 63, MAFRAM 65 has coefficients a and (3 and
functions as a memory, accumulator and filter. The coefficient are DSP
programmable
and controlled by DSP 4. MAFRAM 65 serves the three functions similar to the
AFRAM (61 or 63) but on the magnitude squared data from the AFRAM. As a RAM,
MAFRAM 65 stores 256 correlation lags of data for 1 pilot symbol.
As an example in Figure 6, in the maintenance channel the pilot symbols
are transmitted at the tier 3 rate only. The TDM maintenance channel comprises
two
pilot symbols, with 128 chips per symbol. The four real correlation outputs,
the in-phase
and quadrature phase components (I and Q) of the each user's two pilot symbols
are
stored, accumulated and filtered by AFRAM 61 and AFRAM 63, respectively. The
four
real correlation outputs are denoted IrxIref, Qrxlref, IrxQref and QxxQref.
AFRAM 61
and AFRAM 63 respectively process the two I and Q of two symbols (with 128
chips per
symbol) at the tier 3 rate. In particular, since the pilot symbols are
transmitted in the
maintenance channel at a rate larger than the tier 1 rate (which is the basic
despreading
rate), AFRAM 61 (and similarly AFRAM 63) accumulates the pilot symbols with
filtering. The accumulation is performed by setting the 0 coefficient of AFRAM
61 (and
similarly AFRAM 63) to one. The filtering is accomplished by setting the a and
(3
coefficient of AFRAM 61 (and similarly AFRAM 63) to a value between zero and
one.
Once the frame is processed, AFRAM 61 (and similarly AFRAM 63) overwrites the
old
data and stores/accumulates the pilot symbols of the next user being
transmitted in the
maintenance channel.
After processing in AFRAM 61 and 63, the filtered I and Q (denoted Pcos
and Psin, respectively) are squared at squarers 625 and 627, respectively. The
squared
AFRAM-filtered I and Q components are summed at summer 629, and forwarded to
-16-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
MAFRAM 65 whose output is the post processing signal.
Figure 6A illustrates the a and (3 coefficient settings of MAFRAM 65,
AFRAM 61 and 63 for the tier 1, 2 and 3 data rates. The pilot symbols (denoted
P in the
Data Type row) are shown alongside the data symbols (denoted D) with
corresponding
AFRAM and MAFRAM coefficient settings. The a and 0 coefficient settings
illustratively demonstrate the use of the AFRAMs as a RAM, accumulator and
filter, and
the MAFRAM as a RAM only.
Figure 6B is a lower level diagram illustrating an embodiment of the
AFRAM or MAFRAM according to the invention, e.g., AFRAM 61, 63 or MAFRAM
65 of Figure 6, which is a basic IIR filter. Information is input into a
multiplier 631
where it is multiplied with the a coefficient. RAM 635 is a random access
memory that
stores 256 elements of information. The stored information from RAM 63 5 is
multiplied
with the R coefficient at a multiplier 637. The data multiplied with the gain
coefficients
a and 0 are summed at summer 633 and then forwarded to RAM 635 for storage and
output.
Referring back to Figure 6, AFRAM 61, AFRAM 63, MAFRAM 65,
together with the time-multiplexed correlation engine (e.g., CF core 31),
provide the
ability to search for pilot symbols over a 64-chip window. The pilot symbol
search is a
DSP programmable process for searching the one or more signal components (in
terms
of the received power, signal-to-noise ratio or the multipath width) in the
received
multipath power profile stored in the MAFRAM. The searching can also be
implemented
according to a preprogrammed time alignment. The AFRAM-filtered I and Q values
are
squared and summed for multipath search processing 67 for selecting one or
more signal
components (in terms of the received power, signal-to-noise ratio or multipath
width),
as described herein and in Figures 7 and 7A. The selecting step can also be
implemented
according to a preprogrammed time alignment. Pilot post processor 34 of DSP 4
determines which multipaths are useful for maximum ratio combining to achieve
diversity, which is described in further detail below. DSP 4 then forwards the
best
multipaths to the symbol processors 411 through 418 (Figure 4) in data post
processing
circuit 35 (Figure 3) for pilot symbol aided QPSK demodulation and recovery of
the data
symbols.
-17-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
Figure 7 is a diagram illustrating the search processing of the multipath
power profile according to the invention. Three peaks 1, 2 and 3 are selected
from the
64-chip search window for demodulation. The one or more signal components (in
terms
of the received power, signal-to-noise ratio or the multipath width) are
selected by
processing a succession of passes through the MAFRAM data. The selecting step
can
also be implemented according to a preprogrammed time alignment. The selection
process allows DSP 4 to identify the multipath responses for use in
demodulation and
recovery of the data symbols. In such a selection process, the strongest peak
(or most
desired signal component) in the multipath response of the demodulated
waveform for
the data received from the transmitter is selected and a window (in terms of
time or a
time period) is assigned thereto. The multipath response is examined again,
barring the
window for the strongest peak (peak 1 (or most desired signal component)), and
the
second strongest peak (or second most desired signal component) is selected
and a
window is assigned thereto. The multipath response is examined once more,
barring the
windows for the strongest peak (peak 1 (or most desired signal component)) and
the
second strongest peak (peak 2 (or second most desired signal component)), and
the third
strongest peak (peak 3 (or third most desired signal component)) is selected
and a
window is assigned thereto. The same process steps can be repeated until the
Nth most
desired signal component is selected. Upon completion of the selection
process, the three
peaks 1, 2, 3 (and/or through N) are supplied to digital signal processing DSP
4.
Figure 7A is a flow diagram illustrating an embodiment the multipath
search processing of the multipath response peaks according to the invention.
The
summed squares of AFRAM-filtered I and Q are stored in MAFRAM 65 (Figure 6).
The
information stored in MAFRAM 65 is used for the multipath search processing in
a 64-
chip window as shown in Figure 7. In step 71 of Figure 7A, the strongest peak
(or most
desired signal component) in terms of the received power of the multipath
response is
stored as index 1 and its magnitude stored as max power 1. In step 72, a blank-
out
region 1 is set. The blank-out region is ignored for the next examination of
the multipath
response. In particular, the blank-out region is defined by a lower limit 1
and an upper
limit 1, as follows:
lower limit 1 = index 1 - window (Eq. 2)
-18-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
upper limit 1 = index 1 + window (Eq. 3)
where window denotes the length ofthe multipath response which should not be
searched
on the next pass. In step 73, the strongest peak (or most desired signal
component) in the
multipath response is selected by examining the multipath response, while
ignoring the
blank-out region 1. In effect, the second strongest peak (or second most
desired signal
component) in the entire multipath response is selected, which is stored as
index 2 and
its magnitude stored as max power 2. In step 74, another blank-out region 2 is
set with
a lower limit 2 and an upper limit 2, as follows:
lower limit 2 = index 2 - window (Eq. 4)
upper limit 2 = index 2 + window (Eq. 5)
In step 75, the strongest peak (or most desired signal component) in terms
of the received power (or signal-to-noise ratio, multipath width) in the
multipath response
is selected by examining the multipath response while ignoring the blank-out
regions 1
and 2. The selecting step can also be implemented according to a preprogrammed
time
alignment. In effect, the third strongest peak (or third most desired signal
component)
in the entire multipath response is selected, which is stored as index 3 and
its magnitude
is stored as max power 3. In step 76, an additional blank-out region 3 is set
with a lower
limit 3 and an upper limit 3, as follows:
lower limit 3 = index 3 - window (Eq. 6)
upper limit 3 = index 3 + window (Eq. 7)
In step 77, the noise power of the multipath response is stored. The noise
power is the sum of all the remaining power in the multipath response. In
particular, the
nosie power is obtained by summing all power elements of the multipath
response in the
window, while ignoring blank-out regions 1, 2 and 3. The noise power is then
reported
to DSP 4.
The strongest peaks (or most desired signal component) in terms of the
received power (or signal-to-noise ratio, multipath width) in the multipath
response and
the noise power information in MAFRAM 65 are available to DSP 4 after each
pilot
symbol is processed. With such information, pilot post processor 34 of DSP 4
determines which multipaths are useful for maximum ratio combining to achieve
diversity, which is described in further detail below. DSP 4 then forwards the
best
-19-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
multipaths to the symbol processors 411 through 418 (Figure 4) in data post
processor
35 (Figure 3) for pilot symbol aided QPSK demodulation and recovery of the
data
symbols.
Figure 8 is a diagram illustrating the data post processing in the channels
of the wireless system according to the invention and, more particularly, the
data post
processing in the access channel and the traffic channel. In the data mode of
the access
channel or the traffic channel, data symbols are then processed at the tier I
data rate
which is the basic despreading rate. The basic despreading rate can be 4 or 8
chips per
symbol. No data symbols are transmitted in the maintenance channel.
As the mode controller 41 of DSP 4 (Figure 3) indicates the data mode in
the access channel, data symbols are processed at the tier 1 rate. The four PN
referenced
correlation values, namely Irxlref, Qrxlref, QrxQref and IrxQref are
respectively input
into window processor 43 (Figure 8 and Figure 4) comprising multipliers 811,
812, 813
and 814. Window processor 43 performs a complex de-rotation or multiplication
of the
received correlation values Irxlref, Qrxlref, QrxQref and IrxQref with pilot
symbol
estimates from AFRAM 6,1, 63, and pilot post processor 33. Irxlref and QrxQref
are
multiplied with the AFRAM-filtered in-phase component I (e.g., output Pcos 81
from
AFRAM 61) and QrxIref and IrxQref are multiplied with the AFRAM-filtered
quadrature
phase component Q (e.g., output Psin 83 from AFRAM 63). After de-rotation or
multiplication of the components, the results are input at window select units
801, 802,
803 and 804, respectively, for the multipath selection based on the pilot
search processing
in a 64-chip window as described herein and in Figures 7 and 7A. This pilot
symbol
aided demodulation process produces QPSK outputs which are rotated to a QPSK
signal
constellation. The gate array with the window processor receives the output of
the
correlation filter core and produces a phase-compensated output using a
complex
multiplier for all correlation lags for each 8-chip symbol.
Due to the time multiplexed structure of the correlation core, the 64 chips
.of correlation values associated with each tier 1 data symbol are not in the
proper order
and require time demultiplexing to yield appropriately ordered data symbols.
Upon
completion of the window selection at circuits 801, 802, 803 and 804, the four
real
correlations are combined into I and Q at summers 821 and 823 and
demultiplexed at
-20-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
DEMUX 45.
After demultiplexing, DEMUX 45 outputs the 64 chips of correlation lag
for each tier 1 data symbol to the corresponding symbol processors in circuits
831 and
833. There are eight symbol processors (411 through 418 in circuits 831, 833),
one for
each tier 1 data symbol produced by CF core 31 during the eight phases of the
8-chip
correlation process, described herein and in Figures 4 and 5. The symbol
processors (411
through 418) of one unit 833 take the DSP programmable number of correlation
lags
associated with up to three multipaths and sum the correlation lags to form
three outputs.
The low index (lower limits 1, 2 and/or 3) and high index (upper limits 1, 2
and/or 3) for
all three multipaths determine which correlation lags are used in forming the
three
outputs. DSP 4 performs further integration of the tier 1 data symbols to
provide tier 2
and tier 3 data symbols. This is accomplished by summing four tier 1 data
symbols to
yield one tier 2 data symbol, and summing sixteen tier 1 data symbols to yield
one tier
3 data symbol. DSP 4 then uses the three outputs for combining into a single
output for
signal diversity.
Referring again to Figure 8, DSP 4 further provides post processing of the
data symbols (in data post processor 36) for normalizing the outputs of the
symbol
processors by taking the square root of the absolute value to yield voltage
signals. This
normalization is needed because complex de-rotation or multiplication has been
performed based on the received power (or signal-to-noise ratio, multipath
width) of the
pilot symbols in units of signal power. In order to produce proper diversity
combining
according to maximum ratio combining (MRC) of the three outputs, normalization
is
required for yielding voltage signals. The absolute values (representing the
magnitudes
of the received power) are taken at circuits 841 and 843. Then the square root
of the
signals from absolute value circuits 841, 843 are generated in square root
circuits steps
851 and 853. Next, output of the square root circuits 851, 853 is scaled at
divider 861
and divider 863, respectively. Using MRC as described herein, the three
outputs are then
combined in summing circuits 871 and 873 to obtain one output for each of the
I and Q
components.
Another type of diversity is spatial diversity, where multiple antenna are
provided in the transmitter or receiver for transmitting the same data signals
which
-21-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
provide the diversity needed for combining or selecting the data signals.
Spatial diversity
is achieved by providing the same correlation filter design according to the
invention in
each antenna receiver in the wireless system.
Figure 9 is a flow diagram that illustrates an embodiment of the method
according to,the invention. This method is carried out in a wireless CDMA
system using
a single, common correlation filter (CF). The system has a plurality of
channels with
different data rates and they include the access channel, the maintenance
channel, and the
traffic channel in which information (e.g., pilot or data symbols or both) is
transmitted
at the tier 1, tier 2 and tier 3 rates, as described herein in conjunction
with Figure 2. The
data rate for transmitting the information is programmable by digital signal
processing
(DSP). To begin the method a user-unique code, such as a PN code (or
orthogonal code,
Walsh code), is applied to the information being transmitted in the channels
of the
wireless system (step 93). In step 94, the information is QPSK modulated and
transmitted in any one of the channels. The transmitted information is
correlated at the
basic despreading rate (i.e., the tier 1 rate) using time multiplexing in the
correlation filter
(CF) of the wireless system (step 95). The basic despreading rate can be 4 or
8 chips per
symbol. The correlated information is then demultiplexed (step 96) and QPSK-
demodulated (step 97). The demodulated information is summed (step 98) at the
proper
integer multiple of the tier 1 rate to achieve the tier 2 and tier 3 rates, as
described herein
in conjunction with Figure 4. The one or more signal components (in terms of
the
received power, signal-to-noise ratio or the multipath width) of the multipath
response
of the information are selected in a window or time period for optimal
information
recovery, as described herein in conjunction with Figures 7 and 7A. The
selecting step
can also be implemented according to a preprogrammed time alignment.
Furthermore,
in step 99 three outputs from the demodulated information can be provided and
combined
for temporal diversity, as described herein in conjunction with Figure 8. All
the process
steps according to the invention described herein are advantageously
accomplished using
a single, common correlation filter (CF) which eliminates the need for
additional
correlators or correlation filters for processing received information having
multiple data
rates.
Although the invention has been particularly shown and described in detail
-22-
CA 02400934 2002-08-21
WO 01/63778 PCT/US01/05682
with reference to the preferred embodiments thereof, the embodiments are not
intended
to be exhaustive or to limit the invention to the precise forms disclosed
herein. It will be
understood by those skilled in the art that many modifications in form and
detail may be
made without departing from the spirit and scope of the invention. Similarly,
any process
steps described herein may be interchangeable with other steps to achieve
substantially
the same result. All such modifications are intended to be encompassed within
the scope
of the invention, which is defined by the following claims and their
equivalents.
-23-
